Chloride of Aluminium and its Analysis. 249 



The values of x, x', y, y', are the atomic weights of aluminium 

 as deduced above. They approach to Thomson's number for alu- 

 minium but not to that of Berzelius. 



If the chloride of aluminium analyzed be a sesquichloride, the 

 atomic weight of aluminium calculated as above would be 

 , 5 a; =1.85377 ?/ = 1.82350 



" ^ x'= 1 .85430 y = 1 .82274 



Tluis it seems that the atomic number calculated for aluminium, 

 upon, the supposition that the chloride contains 1 al. + lcl., is smaller 

 than Thomson's, and 1 al. + lj cl. is larger than that of Berzelius. 

 If the atomic weight of ahunina be determined, we shall from that 

 be enabled to determine whether the chloride of aluminium be a 

 chloride orsesquichloride. Berzelius says that alumina is composed of 

 Aluminium, - 53.3 

 Oxygen, - - 46.7 



*100.0 

 Let us see how this corresponds with the analysis of chloride of 

 aluminium. 



Alumina. Aluminium. Alumina. Aluminium. 



C 100 : 53.3 :: 0.2975t : x and x= 0.158567 



I < Alninina. C'xv?pn. Oxv<ren. 



( 100 : 46.7 :: 0.2975t : y and 3/ = 0.138932 

 But the aluminium obtained was 0.139188t, and the oxygen in 

 the alumina was 0.15831 18*^. These numbers deduced from exper- 

 iment, correspond very closely with those deduced from Berzelius' 

 composition of alumina, but they are inverse to each other. 



I infer from this, that the numbers as laid down in Berzelius' 

 Traite de Chimie for oxygen and aluminium in the composition of 

 alumina, were accidentally transposed in the original manuscripts of 

 his experiments. f I have no doubt that the composition of alumina 

 should have been written, 



♦ Traill de Chimie, Tome II. p. 373. t Vide table a. 



S 0.1583118 oxygen =0.2975 alnmin.-) —0.139188 aluDiininm. 



t The same kind of error ofcnrs in HaUy's Tiaii^ de Mineralogie, 2d Edition, 

 Tunie II. p. 214. The angles ol the ilioiiihoid of niirnte of s.)da are there laid 

 down P on P'orP' 106° Iti' 



p " p' " p" 73° W 



In measuring some fine c;y.stal.«;of this salt in 1830. hythe reflective goniometer 

 I found the angles 10 be P on F' oi P" UX;° M' 

 p " |,' '• [," 73° W 



The miniites in ITndy are inverse to those observed. \ do not K-now tint this er- 

 r< r has been observed or corrected. 



Vol. XXVIl.— No. 2. 32 



